Search

Many places around the world are affected by rainfall-induced landslides. These types of landslides often occur in unsaturated soil slopes, and conventional soil mechanics does not commonly consider the effects of negative pore- pressure, or suction, in effective stress calculations for slope stability. The soil water characteristic curve (SWCC) defines the relationship between moisture content and suction in a soil due to interaction between the air and water phases in a saturating or desaturating soil, and can be used to derive relationships for unsaturated soil permeability and water storage. This relationship has been found pivotal in describing unsaturated soil behaviour and was explored in detail in this thesis. Rainfall-induced landslides frequently occur in remote or difficult to access locations, which still remain capable of impacting human life, infrastructure and the environment. These areas can be monitored with sensors connected via the Internet of Things to a local server. Moisture content and suction sensors were purchased by NTNU for this work with the purpose of assessing measurement accuracy and reliability in a laboratory setting, and the possibility of developing an in-situ SWCC during infiltration testing. During this thesis a large-scale infiltration column was designed and constructed, standing 1.3 m tall with 0.24 m internal diameter. Material selection was conducted prior to delivery of the instruments and completion of the large- scale column through initial testing consisting of small-scale infiltration testing and numerical analysis using PLAXIS. The results of the initial testing concluded a combination of three materials available at NTNU was required to obtain a soil suction range compatible with the measurement range of the ordered suction sensor while maintaining reasonable infiltration times. The combined material contained 45% sand fraction, 45% silt fraction and 10% clay fraction and was classified as a clayey, sandy silt according to Norwegian standards. The infiltration column was filled with 1 m of unsaturated soil and a constant head of water was applied to the top surface until the column became fully saturated through infiltration. Saturated permeabilities were measured from the flow of water exiting the base of the column. Five pairs of sensors were installed at various depths in the column to monitor the suction and moisture content change with time during infiltration. The sensors were connected to a computer through dataloggers, and a software program was written to record sensor readings at set intervals. Three column infiltration tests were completed during this semester. The sensor data collected was used to create an average SWCC for the tested material using the van Genuchten-Mualem curve fitting model. From the SWCC, three methods were used to estimate the unsaturated Soil Permeability Function (SPF) for the material, using a combination of statistical methods, instrument data and visual measurements. The derived SPFs gave differing results due to assumptions made in each method definition, but gave a trend within each infiltration test. An attempt was made to calibrate a Green-Ampt infiltration model to the infiltration rate, however the material did not fit with established datasets for obtaining the suction at the wetting front. The finite element software PLAXIS 2D was used for numerical analysis of the infiltration tests to evaluate laboratory infiltration results. The derived SWCC was input as a hydraulic property of the material with the obtained saturated permeability, and a flow only analysis with transient groundwater flow conducted. The infiltration times were overestimated by 2.6 to 4 times that of the column test results. A sensitivity analysis was run on the van Genuchten- Mualem curve fitting parameters "a" and "n" to determine the impact on the infiltration time by modification of the parameters. The analysis concluded slight modifications to the SWCC can have large impacts on infiltration time, and due to the scatter in the sensor data it is difficult to estimate the appropriate SWCC for the material to match numerical analysis results. The main conclusion from this thesis is the suction sensor ordered was not appropriate for the tested material. The sensor is not capable of measuring suctions below 9 kPa, which in this thesis meant the air entry value for the test material could not be found through sensor measurements. The transitional zone of the SWCC was also on the lower boundary of the suction sensor range and showed large scatter. Soils which desaturate more slowly and at higher suctions could be measured with this sensor, however infiltration times in a 1 m tall clay column would take significant time and may not be suitable for master thesis work. Additionally, the suction sensor appeared to have a slow reaction time to quickly changing moisture contents, which was shown by SWCCs differing between higher and lower sensor pairs as the infiltration rate slowed with depth. Additional work could be conducted on the response time of the sensors during infiltration to conclude if they are suitable for accurate monitoring of infiltration in a field setting. This thesis was completed in conjunction with the KlimaDigital project, a collaboration between NTNU and SINTEF, among other industry and public partners (SINTEF). The KlimaDigital project aims to create a digitally supported framework to assess geohazard risks around Norway, and is currently focusing on rainfall-induced landslides. Based on the results of this thesis, the instruments may be installed in a field setting around Trondheim in the summer for further evaluation.